def by_chromosome(context, chromosome):
vm = context.vmf.read_row_group(chromosome - 1).to_pandas()
if args.frequency_filter:
vm = filter_by_frequency(vm, args.frequency_filter)
g = context.get_genotype_file(chromosome)
regions = context.regions
regions = regions[regions.chr == "chr{}".format(chromosome)]
for i, region in enumerate(regions.itertuples()):
logging.log(9, "Processing region in chr %d: %d/%d", chromosome, i + 1,
regions.shape[0])
vmw = Genomics.entries_for_window(chromosome,
region.start - args.window,
region.stop + args.window, vm)
ids = vmw.id.values
logging.log(9, "%d variants", len(ids))
d = Parquet._read(g, columns=ids, skip_individuals=True)
d = numpy.array([d[x] for x in ids], dtype=numpy.float32)
if context.args.standardise_geno:
cov = numpy.corrcoef(d, ddof=1).astype(numpy.float32, copy=False)
else:
cov = numpy.cov(d).astype(numpy.float32, copy=False)
logging.log(9, "%d rows", cov.shape[0])
context.sink(cov, ids, region)
def count_variants(chromosome, start, end, vf, m, last_chromosome, args):
try:
chromosome = int(chromosome.split("chr")[1])
start = int(start)
end = int(end)
if chromosome != last_chromosome:
logging.info("Reading chromosome %d", chromosome)
m = vf.read_row_group(chromosome - 1).to_pandas()
last_chromosome = chromosome
if args.frequency_filter:
logging.log(9, "Filtering by frequency")
m = m[(m.allele_1_frequency > args.frequency_filter)
& (m.allele_1_frequency < 1 - args.frequency_filter)]
v = Genomics.entries_for_window(chromosome, start, end, m)
count = v.shape[0]
except:
count = "NA"
return count, m, last_chromosome
